Machine-learned disambiguation of user action data

1. A computer-implemented method comprising: receiving data relating a user to entities of a plurality of entity types, wherein the data includes references to entities that are not names of the entities; annotating the data using a machine-learning based annotator that is trained to recognize entities and entity attributes of the entities in the data, the annotating by the machine-learning based annotator including: identifying an entity reference in the data that is not a name of an entity, identifying multiple candidate entities for the entity reference based in part on the entity reference, selecting a particular entity of the multiple candidate entities based in part on information in a user profile of the user, and annotating the entity reference in the data with an unambiguous entity identifier for the particular entity; receiving a query from the user, wherein the query includes one or more terms that specify one or more entity identifiers indicative of a first entity type and a particular relation of the user with respect to the entities; determining entities that are of the first entity type; querying the annotated data for data indicating the particular relation of the user with respect to the entities of the first entity type; generating an answer to the query from the annotated data indicating the particular relation of the user with respect to the entities of the first entity type; and providing the answer to the query to the user.

2. The method of claim 1 , wherein the query is received from the user by voice input.

3. The method of claim 1 , wherein the query is received from the user by keyboard, mouse, or touchpad inputs.

4. The method of claim 1 , further comprising: associating at least one identified entity with the user profile by annotating the user profile with the identified entity; and customizing an experience of the user using the annotated user profile.

5. The method of claim 1 , wherein the machine-learning based annotator selects the particular entity based at least on content in the data that represents actions taken by the user.

6. The method of claim 1 , wherein generating the answer further comprises aggregating results based on a date or category.

7. A system comprising: one or more computers; and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising: receiving, by the system, data relating a user to entities of a plurality of entity types, wherein the data includes references to entities that are not names of the entities; annotating the data using a machine-learning based annotator that is trained to recognize entities and entity attributes of the entities in the data, the annotating by the machine-learning based annotator including: identifying, by the system, an entity reference in the data that is not a name of an entity, identifying, by the system, multiple candidate entities for the entity reference based in part on the entity reference, selecting, by the system, a particular entity of the multiple candidate entities based in part on information in a user profile of the user, and annotating, by the system, the entity reference in the data with an unambiguous entity identifier for the particular entity; receiving, by the system, a query from the user, wherein the query includes one or more terms that specify one or more entity identifiers indicative of a first entity type and a particular relation of the user with respect to the entities; determining, by the system, entities that are of the first entity type; querying, by the system, the annotated data for data indicating the particular relation of the user with respect to the entities of the first entity type; generating, by the system, an answer to the query from the annotated data indicating the particular relation of the user with respect to the entities of the first entity type; and providing, by the system, the answer to the query to the user.

8. The system of claim 7 , wherein the query from the user is received from the user by voice input.

9. The system of claim 7 , wherein the query from the user is received by the system from the user by keyboard, mouse, or touchpad input.

10. The system of claim 7 , wherein the operations further comprise: associating at least one identified entity with a user profile by annotating the user profile with the identified entity; and customizing an experience of the user using the annotated user profile.

11. The system of claim 7 , wherein the machine-learning based annotator selects the particular entity based at least on content in the data that represents actions taken by the user.

12. The system of claim 7 , wherein generating the answer further comprises aggregating results based on a date or a category.

13. At least one non-transitory computer-readable storage medium comprising instructions stored thereon that are executable by a processing device and upon such execution cause the processing device to perform operations comprising: receiving data relating a user to entities of a plurality of entity types, wherein the data includes references to entities that are not names of the entities; annotating the data using a machine-learning based annotator that is trained to recognize entities and entity attributes of the entities in the data, the annotating by the machine-learning based annotator including: identifying an entity reference in the data that is not a name of an entity, identifying multiple candidate entities for the entity reference based in part on the entity reference, selecting a particular entity of the multiple candidate entities based in part on information in a user profile of the user, and annotating the entity reference in the data with an unambiguous entity identifier for the particular entity; receiving a query from the user, wherein the query includes one or more terms that specify one or more entity identifiers indicative of a first entity type and a particular relation of the user with respect to the entities; determining entities that are of the first entity type; querying the annotated data for data indicating the particular relation of the user with respect to the entities of the first entity type; generating an answer to the query from the annotated data indicating the particular relation of the user with respect to the entities of the first entity type; and providing the answer to the query to the user.

14. The at least one non-transitory computer-readable storage medium of claim 13 , wherein the entity reference is text that is different than the name of the entity.

15. The at least one non-transitory computer-readable storage medium of claim 14 , wherein the text is used as a pattern for identifying the multiple candidate entities.

16. The at least one non-transitory computer-readable storage medium of claim 13 , wherein the machine-learning based annotator selects the particular entity based at least on content in the data that represents actions taken by the user.

17. The at least one non-transitory computer-readable storage medium of claim 13 , wherein the annotating by the machine-learning based annotator further includes: identifying in the data candidate entity attributes; and using the candidate entity attributes in selecting the particular entity.

18. The at least one non-transitory computer-readable storage medium of claim 13 , wherein the operations further comprise associating the particular entity the user profile by annotating the user profile with the particular entity.

19. The at least one non-transitory computer-readable storage medium of claim 18 , wherein the operations further comprise customizing an experience of the user using the annotated user profile.

20. The at least one non-transitory computer-readable storage medium of claim 13 , wherein the information in the user profile used to select the particular entity includes location data.